|Publication number||US7302652 B2|
|Application number||US 10/404,937|
|Publication date||Nov 27, 2007|
|Filing date||Mar 31, 2003|
|Priority date||Mar 31, 2003|
|Also published as||US20040194037|
|Publication number||10404937, 404937, US 7302652 B2, US 7302652B2, US-B2-7302652, US7302652 B2, US7302652B2|
|Inventors||Zhanping Chen, Lakshman Thiruvenkatachari, Shahram Jamshidi|
|Original Assignee||Intel Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (4), Non-Patent Citations (6), Referenced by (7), Classifications (5), Legal Events (7)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present invention concerns methods, systems, circuits, and software for controlling leakage current in integrated circuits.
In recent years, the popularity of battery-powered electronic devices, such as laptop computers, personal digital assistants, and cellular telephones, has grown dramatically. This growth, in turn, has fueled consumer demand and expectations for longer battery life, and driven manufacturers and researchers to focus more attention on improving the energy efficiency of the microprocessors and other integrated circuits that enable these devices.
Integrated circuits, also known as “chips,” are interconnected networks of electrical components, fabricated on a common foundation, or substrate, of semiconductor material. These circuits typically comprise millions of microscopic transistors. A key aspect of energy efficiency in integrated circuits is the control of leakage current in these transistors.
Leakage current refers to electric current that a transistor conducts when turned off. Ideally, this current is zero; however, in practice, all transistors exhibit some level of leakage current. (Leakage current is analogous to water that flows from a leaky faucet.) The cumulative leakage for a circuit having millions of transistors can amount to a significant amount of wasted power—known as leakage power. For example, in some circuits, leakage power may account for as much as one third of total power usage.
Although there are a number of techniques available to reduce leakage current, there is still considerable room for improvement. For example, one prevailing technique is vector control, which entails applying a single, optimized input vector (that is, a particular set of input signals) to an entire integrated circuit to lock its transistors in a collectively reduced or optimal leakage state. However, in studying this technique, the present inventors have recognized that it becomes increasingly ineffective as circuit complexity or size increases.
Accordingly, there is a need for better ways of reducing leakage current in integrated circuits, particularly larger, complex circuits, such as microprocessors.
This description, which references and incorporates the above-identified figures and the appended claims, describes one or more specific embodiments of one or more inventions. These embodiments, offered not to limit but only to exemplify and teach the one or more inventions, are shown and described in sufficient detail to enable those skilled in the art to implement or practice the invention. Thus, where appropriate to avoid obscuring the invention, the description may omit certain information known to those of skill in the art.
Specifically, circuit 110 includes: N circuit portions, of which portions 112, 114, 116, and 118 are representative; X primary (or circuit boundary) inputs, of which IN1, IN2, IN3, and INX, are representative; and Y primary outputs, of which OUT1, OUT2, and OUTY are representative.
Circuit portions 112-118 generally include any portion of circuitry in the integrated circuit. In the exemplary embodiment, which concerns a complex digital CMOS (complementary metal-oxide-semiconductor) circuit, such as a microprocessor, each portion includes non-critical-path components (not shown separately) and embodies one or more forms of leakage-reduction technology. Exemplary forms of leakage-reduction technology include multiple-supply voltage (Vcc) CMOS technology, multiple-threshold CMOS, sleep-transistor structure, and reverse-body bias CMOS. In some embodiments, one or more of the circuit portions embody all of these leakage-reduction technologies; in others, various circuit portions embody combinations of one, two, or three of these technologies.
Additionally, circuit portions 112-118 include respective multiplexers 112.1-118.1, with each multiplexer having input sets A and B, an output set C, and a select input S, with each input set having of one or more inputs, and each output set having one or more outputs. (Multiplexers 112-118 may have different numbers of inputs and outputs.) Input sets A for multiplexers 112.1, 114.1, and 116.1 are coupled respectively to primary inputs IN1, IN2, and INX, and input set A for multiplexer 118.1 is coupled to an internal output of circuit portion 116. Output sets C for multiplexers 112.1, 114.1, 116.1, and 118.1 are coupled to input nodes (not shown) of their respective circuit portions. Select input S for each multiplexer selectively couples its set A or set B inputs to its set C outputs. In the exemplary embodiment, select input S is coupled to primary input IN3, which receives a clock-enable signal. The clock-enable signal, at least in a microprocessor context, is indicative of a standby or power-conservation mode; some other embodiments may couple the select input to a gated clock signal or other internal or external control signal. Set B inputs for the multiplexers are coupled to data-storage device 120.
Data-storage device 120 includes a machine-readable medium, such as a volatile or non-volatile memory. In the exemplary embodiment, device 120 includes a non-volatile memory positioned on an integrated-circuit chip with circuit 10. However, in other embodiments, device 120 is positioned on a separate integrated circuit or data-storage apparatus. Data-storage device 120 includes leakage-control data structure 130.
Leakage-control data structure 130 includes a set of one or more leakage-control vectors (LCVs), such as LCVs 132, 134, 136, and 138. Each LCV corresponds to one of the circuit portions and includes a set of binary input values selected to establish a standby leakage current level for its corresponding circuit portion. More precisely, LCVs 132, 134, 136, and 138 include respective sets of binary input values for establishing standby leakage currents for respective circuit portions 112, 114, 116, and 118. In the exemplary embodiment, each LCV is defined to minimize or at least reduce the standby leakage current for its corresponding circuit portion relative to the leakage that would occur with other input vectors. In some embodiments, one or more of the LCVs may be applicable to more than one circuit portion. The LCVs can be generated randomly, by enumeration, by adaptive algorithm, such as a genetic algorithm, or by some heuristic.
General operation of system 100 entails integrated circuit 110 receiving a command, such as standby-mode or sleep-mode command, from an operating system, power-management system, or other command-issuing component of a mobile device (not shown in this figure). In response to the command, which can, for example, take the form of a clock-disabling signal on input IN3, integrated circuit 110 couples the input nodes of each of circuit portions 112-118 to input set B of its corresponding multiplexer and thus its corresponding LCV in data-storage device 120. (In some embodiments, the LCVs may be effectively hardwired into the integrated circuit by coupling the set B inputs of the multiplexers to appropriate logic voltage levels, such as upper and lower power supply nodes, in the integrated circuit rather than to a data-bearing memory structure inside or outside the circuit. As an example,
As a result of applying the LCVs, circuit portions 112-118 enter a low- or reduced leakage state based on the applied LCVs. The LCVs remain in effect until another command, such as a clock-enabling signal or other control signal deselects the set B inputs of the multiplexer. In some embodiments, the multiplexers may include multiple select inputs to allow selection and application of other sets of specialized input vectors to the circuit blocks, for example, to warm, restart, or otherwise prepare the circuit for continued activity.
Exemplary execution begins at block 310, which entails input or receipt of a circuit definition or specification. In the exemplary embodiment, this definition takes the form of a net listing. Execution then advances to block 320.
Block 320 entails partitioning the circuit into two or more portions or clusters. For circuits with pipeline structure, the exemplary embodiment partitions the pipeline structure at the sampling elements, such as flip-flops or latches, between the various circuit stages as shown in exemplary leakage-reduction system 400 in
More particularly, system 400 includes an integrated circuit 410 and a leakage-control data structure 420. Integrated circuit 410, which in some embodiments constitutes a microprocessor or digital signal processor, includes N pipelined circuit stages of which circuit stages 411, 412, and 413 are representative. Circuit stages (or blocks) 411-413 are driven by respective sampling elements 414, 415, and 416. In the exemplary embodiment, each of these sampling elements includes a latch or a flip-flop.
Leakage-control data structure 420 includes a set of one or more LCVs, such as LCVs 421, 422, 423, which are respectively associated with flip-flop stages 414, 415, and 416. The pipeline-based partitioning illustrated in
Block 340 determines whether LCVs for all the clusters defined at block 320 have been determined. If there are clusters that lack a corresponding LCV, then execution returns to block 350 to determine an LCV for another one of the defined clusters. However, if each of the defined clusters has a corresponding LCVs, execution continues to block 350.
Block 350 entails outputting the LCVs to a data-storage device, such as device 120 in
Block 502 entails receiving a circuit definition having primary inputs and outputs (or more generally boundary nodes). In the exemplary embodiment, the primary inputs include the data and address pins of the circuit, and the primary outputs include other pins that output data or otherwise indicate a boundary of the circuit. However, in other embodiments, any input pin or node may be treated as a primary input. Some embodiments may define the primary inputs and outputs to effectively confine or focus activities of the leakage-control module to specific areas of a circuit definition, such as non-critical path areas. Exemplary execution continues at block 504.
Block 504 entails defining an input queue including the primary inputs. In the exemplary embodiments, the Q is arranged such that the primary inputs are arranged in an order corresponding to their arrangement on a pin-out diagram of the circuit. However, some other embodiments use other input ordering. Execution then advances to block 506.
Block 506 determines if the input queue is empty or not. If the input queue is determined to be empty, execution branches to block 508, which entails outputting results of the exemplary method in the form of new circuit definition and a set of corresponding LCVs. (Some embodiments output the LCV and the new circuit definition cluster by cluster after the acceptance of each cluster at block 518 and before execution of block 520.) However, if the input queue is determined not to be empty, execution advances to block 510.
Block 510 entails defining a temporary circuit cluster. In the exemplary embodiment, this entails selecting an input from the queue, for example, the next available input; searching the circuit definition for any subcircuits or circuit blocks, such as logic gates, driven by the selected input. (Some embodiments select two or more inputs at a time, such as two or more adjacent inputs in the queue.) If any outputs of the found circuit blocks are not primary outputs of the original circuit, the exemplary embodiment adds one or more cluster-boundary devices, such as multiplexers, flip-flops, latches, or other data-sampling elements, between each of these non-primary outputs and the inputs of any circuit blocks it drives, to define a temporary cluster. Execution proceeds to block 512.
Block 512 entails determining a leakage-control vector for the temporary cluster. The exemplary embodiment determines an optimal leakage-control vector for the temporary cluster by random testing, by enumeration, by adaptive algorithm, such as a genetic algorithm, or by some heuristic. The leakage-control vector for the temporary cluster is associated with a temporary leakage value defined as the temporary-best-leakage (tmpBestLkg) for the original circuitry of the temporary cluster and an extra-leakage value (extraLkg) for the added boundary devices. The leakage values can be determined using a simulation program or other technique, such as equation-based evaluation.
Block 514 entails determining whether to expand the temporary cluster. In the exemplary embodiment, this determination entails determining whether the temporary cluster meets the following criterion:
where extraLkg denotes leakage of the temporary cluster attributable to the added cluster-boundary device (and supporting circuitry); t % denotes the targeted reduction percentage, for example −5, −10, −15, −20, or −25 percent; and avgLkg denotes the average leakage of the temporary cluster. The exemplary embodiment defines the average leakage as the cumulative leakage of the temporary cluster for a number of input vectors divided by the number of input vectors. Another embodiment defines the average leakage for the cluster as the number of gates or circuit blocks in the original circuit times the ratio of the total leakage for the original circuit to the number of gates (or circuit blocks) in the original circuit. Still other embodiments may use other measures of central tendency to define appropriate cluster-growth or -selection criteria. Other embodiments may define leakage-based, cluster-shrinkage criteria that recursively or iteratively shrinks from larger temporary clusters down toward smaller optimal cluster sizes, by for example, determining whether the leakage for the current temporary cluster is less than that for the previous temporary cluster, before further shrinking the cluster.)
Block 516, which follows a determination at block 514 to expand the cluster, entails adding more circuitry to the current temporary cluster. In the exemplary embodiment, this entails copying the current temporary cluster to a previous temporary cluster, removing any previously added boundary devices, and then determining whether any of the outputs of the temporary cluster (minus the previously added boundary devices) are non-primary outputs. If this cluster has any non-primary outputs, the exemplary method adds any circuit blocks driven by these non-primary outputs to the cluster along with corresponding cluster-boundary devices to any non-primary outputs for these added circuit blocks, thereby defining a new temporary cluster. (Other embodiments need not expand the cluster by adding circuit blocks that are driven by non-primary outputs. For example, some embodiments may expand the cluster by adding one or more other adjacent or even non-adjacent inputs from the queue along with circuit blocks driven by these added inputs. Still other embodiments may add circuit blocks without regard for their input connections.)
If, however, the current temporary cluster has no non-primary outputs (that is, it has only primary outputs), then a primary input, such as the next available primary input, is selected from the input queue, and added to the current temporary cluster, along with any circuit blocks driven by this added input and any corresponding cluster-boundary devices. Execution then returns to block 512 to determine whether to further expand the cluster. If the new temporary cluster is unacceptable, indicating that the temporary cluster has grown too large, then execution advances to block 518.
Block 518 entails accepting a defined cluster. In the exemplary embodiment, the cluster that triggers execution of block 518 is actually one-iteration too large; so, acceptance entails storing the previous temporary cluster to a list or file of permanent cluster definitions for the original circuit definition. (Other embodiments may accept other defined clusters based on the structure of the expansion criteria.) Execution then continues at block 520.
Block 520 updates the input queue based on the accepted cluster. To this end, the exemplary embodiment adds any non-primary outputs of this accepted cluster to, for example, the front or the back of, the input queue created at block 504, and clears or restores any stored variables of the current temporary and previous temporary clusters. Execution then returns to block to 506.
Memory system 640, which can include any form of volatile or non-volatile data-storage technology, such as electric, ferroelectric, magnetic, or optical, includes standby power software 642. Processing unit 650 includes a microprocessor, digital-signal processor, and/or other integrated circuit, with at least one of these components including a leakage-reduction system, such as one corresponding to system 100 or 400 or a related embodiment described above.
Accessory 660 includes interface circuitry and related connectors for adding detachable modules to system 600. Exemplary modules include mobile telephone transceivers, network communicators, memory extensions, infrared transceivers, digital cameras, barcode readers, digital media players, etc. In some embodiments, these modules are permanently integrated into accessory 660 and thus form a permanent part of system 600. Additionally, one or more components of accessory 660 may include a leakage-reduction system corresponding to system 100 or 400 or a related embodiment described above.
The embodiments described in this document are intended only to illustrate and teach one or more ways of practicing or implementing the present invention, not to restrict its breadth or scope. The actual scope of the invention, which embraces all ways of practicing or implementing the teachings of the invention, is defined only by the following claims and their equivalents.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US6014751 *||May 5, 1997||Jan 11, 2000||Intel Corporation||Method and apparatus for maintaining cache coherency in an integrated circuit operating in a low power state|
|US6191606 *||Sep 10, 1998||Feb 20, 2001||Intel Corporation||Method and apparatus for reducing standby leakage current using input vector activation|
|US6493856 *||Jun 29, 2001||Dec 10, 2002||Kabushiki Kaisha Toshiba||Automatic circuit generation apparatus and method, and computer program product for executing the method|
|US6687883 *||Dec 28, 2000||Feb 3, 2004||International Business Machines Corporation||System and method for inserting leakage reduction control in logic circuits|
|1||Chen, Z , et al., "IDDQ Testing for Deep Submicron ICs: Challenges and Solutions", IEEE Design and Test of Computers, 19, (2002), 24-33.|
|2||Chen, Zhanping , et al., "Estimation of standby leakage power in CMOS circuits considering accurate modeling of transistor stacks", Proceedings of the International Symposium on Low Power Design, (Aug. 10-12, 1998), 239-244.|
|3||Goldberg, D. E., "Genetic Algorithms in Search, Optimization & Machine Learning", Addison-Wesley, Reading, MA, Chapters 1, 4, 5, 7 and 8, (1989), 1-25, 89-216, and 261-312.|
|4||Holland, John H., "Adaptation in Natural and Artificial Systems", University of Michigan Press, Ann Arbor, MI, Chapters 3, 8 and 9, (1975), 32-65 and 141-170.|
|5||Roy, K , et al., "Multiple Vdd Multiple Vth (MVCMOS) for Lower Power Applications", 1999 International Symposium on Circuits and Systems, 1, (1999), 366-370.|
|6||Wei, Liqiong , et al., "Design and optimization of low voltage high performance dual threshold CMOS circuits", Proceedings-Design Automation Conference, (Jun. 15-19, 1998), 489-494.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7592840||Nov 8, 2006||Sep 22, 2009||Intel Corporation||Domino circuit with disable feature|
|US8736332||Dec 17, 2009||May 27, 2014||Lsi Corporation||Leakage current reduction in a sequential circuit|
|US9553581 *||Nov 1, 2015||Jan 24, 2017||Freescale Semiconductor, Inc.||Package-aware state-based leakage power reduction|
|US20070168792 *||Dec 4, 2006||Jul 19, 2007||International Business Machines Corporation||Method to Reduce Leakage Within a Sequential Network and Latch Circuit|
|US20080106302 *||Nov 8, 2006||May 8, 2008||Kin Yip Sit||Domino circuit with disable feature|
|US20110148496 *||Dec 17, 2009||Jun 23, 2011||Srinivas Sriadibhatla||Leakage current reduction in a sequential circuit|
|EP2339752A1||Mar 30, 2010||Jun 29, 2011||LSI Corporation||Leakage current reduction in a sequential circuit|
|Cooperative Classification||G06F2217/08, G06F17/5045|
|Mar 31, 2003||AS||Assignment|
Owner name: INTEL CORPORATION, CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:CHEN, ZHANPING;THIRUVENKATACHARI, LAKSHMAN;JAMSHIDI, SHAHRAM;REEL/FRAME:013934/0285
Effective date: 20030331
|Apr 22, 2008||CC||Certificate of correction|
|May 18, 2011||FPAY||Fee payment|
Year of fee payment: 4
|Jul 4, 2013||AS||Assignment|
Owner name: MICRON TECHNOLOGY, INC., IDAHO
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:INTEL CORPORATION;REEL/FRAME:030747/0001
Effective date: 20111122
|May 20, 2015||FPAY||Fee payment|
Year of fee payment: 8
|May 12, 2016||AS||Assignment|
Owner name: U.S. BANK NATIONAL ASSOCIATION, AS COLLATERAL AGEN
Free format text: SECURITY INTEREST;ASSIGNOR:MICRON TECHNOLOGY, INC.;REEL/FRAME:038669/0001
Effective date: 20160426
|Jun 2, 2016||AS||Assignment|
Owner name: MORGAN STANLEY SENIOR FUNDING, INC., AS COLLATERAL
Free format text: PATENT SECURITY AGREEMENT;ASSIGNOR:MICRON TECHNOLOGY, INC.;REEL/FRAME:038954/0001
Effective date: 20160426